- Fix a crash when pausing a subscription to
TarEntry.contents right before it ends.
- Fix generating corrupt tar files when adding lots of entries at very high speeds (#20).
- Allow tar files with invalid utf8 content in PAX header values if those values aren't used for anything important.
- Improve error messages when reading a tar entry after, or during, a call to
- Improve performance when reading large archives
- Support sync encoding with
- Support generating tar files with GNU-style long link names
format parameter to
chunked_stream dependency in favor of
- Allow arbitrarily many zero bytes at the end of an archive when
disallowTrailingData is enabled.
disallowTrailingData parameter to
TarReader. When the option is set,
readNext will ensure that the input stream does not emit further data after the tar archive has been read fully.
- Remove outdated references in the documentation
- Fix some minor implementation details
Most of the tar package has been rewritten, it's now based on the implementation written by Garett Tok Ern Liang in the GSoC 2020.
tar prefix to exported symbols.
TarEntry.data to create a tar entry from bytes.
WritingSink private. Use
tarWritingSink to create a general
TarReader is now a
StreamIterator, the transformer had some design flaws.
- Support writing user and group names
- Better support for PAX-headers and large files